2 * Jalview - A Sequence Alignment Editor and Viewer ($$Version-Rel$$)
3 * Copyright (C) $$Year-Rel$$ The Jalview Authors
5 * This file is part of Jalview.
7 * Jalview is free software: you can redistribute it and/or
8 * modify it under the terms of the GNU General Public License
9 * as published by the Free Software Foundation, either version 3
10 * of the License, or (at your option) any later version.
12 * Jalview is distributed in the hope that it will be useful, but
13 * WITHOUT ANY WARRANTY; without even the implied warranty
14 * of MERCHANTABILITY or FITNESS FOR A PARTICULAR
15 * PURPOSE. See the GNU General Public License for more details.
17 * You should have received a copy of the GNU General Public License
18 * along with Jalview. If not, see <http://www.gnu.org/licenses/>.
19 * The Jalview Authors are detailed in the 'AUTHORS' file.
21 package jalview.analysis.scoremodels;
23 import jalview.api.analysis.PairwiseScoreModelI;
24 import jalview.api.analysis.SimilarityParamsI;
25 import jalview.api.analysis.SimilarityScoreModelI;
26 import jalview.datamodel.AlignmentView;
27 import jalview.math.Matrix;
28 import jalview.math.MatrixI;
29 import jalview.util.Comparison;
31 import java.util.Arrays;
34 * A class that models a substitution score matrix for any given alphabet of
37 public class ScoreMatrix implements SimilarityScoreModelI,
41 * this fields records which gap character (if any) is used in the alphabet;
42 * space, dash or dot are recognised as gap symbols
44 private char gapCharacter = '0';
47 * Jalview 2.10.1 treated gaps as X (peptide) or N (nucleotide)
48 * for pairwise scoring; 2.10.2 uses gap score (last column) in
49 * score matrix (JAL-2397)
50 * Set this flag to true (via Groovy) for 2.10.1 behaviour
52 private static boolean scoreGapAsAny = false;
54 public static final short UNMAPPED = (short) -1;
56 private static final String BAD_ASCII_ERROR = "Unexpected character %s in getPairwiseScore";
58 private static final int MAX_ASCII = 127;
61 * the name of the model as shown in menus
62 * each score model in use should have a unique name
67 * a description for the model as shown in tooltips
69 private String description;
72 * the characters that the model provides scores for
74 private char[] symbols;
77 * the score matrix; both dimensions must equal the number of symbols
78 * matrix[i][j] is the substitution score for replacing symbols[i] with symbols[j]
80 private float[][] matrix;
83 * quick lookup to convert from an ascii character value to the index
84 * of the corresponding symbol in the score matrix
86 private short[] symbolIndex;
89 * true for Protein Score matrix, false for dna score matrix
91 private boolean peptide;
93 private float minValue;
95 private float maxValue;
98 * Constructor given a name, symbol alphabet, and matrix of scores for pairs
99 * of symbols. The matrix should be square and of the same size as the
100 * alphabet, for example 20x20 for a 20 symbol alphabet.
103 * Unique, human readable name for the matrix
105 * the symbols to which scores apply
107 * Pairwise scores indexed according to the symbol alphabet
109 public ScoreMatrix(String theName, char[] alphabet, float[][] values)
111 if (alphabet.length != values.length)
113 throw new IllegalArgumentException(
114 "score matrix size must match alphabet size");
116 for (float[] row : values)
118 if (row.length != alphabet.length)
120 throw new IllegalArgumentException(
121 "score matrix size must be square");
125 this.matrix = values;
127 this.symbols = alphabet;
129 symbolIndex = buildSymbolIndex(alphabet);
134 * crude heuristic for now...
136 peptide = alphabet.length >= 20;
140 * Record the minimum and maximum score values
142 protected void findMinMax()
144 float min = Float.MAX_VALUE;
145 float max = -Float.MAX_VALUE;
148 for (float[] row : matrix)
154 min = Math.min(min, f);
155 max = Math.max(max, f);
165 * Returns an array A where A[i] is the position in the alphabet array of the
166 * character whose value is i. For example if the alphabet is { 'A', 'D', 'X'
167 * } then A['D'] = A[68] = 1.
169 * Unmapped characters (not in the alphabet) get an index of -1.
171 * Mappings are added automatically for lower case symbols (for non case
172 * sensitive scoring), unless they are explicitly present in the alphabet (are
173 * scored separately in the score matrix).
175 * the gap character (space, dash or dot) included in the alphabet (if any) is
176 * recorded in a field
181 short[] buildSymbolIndex(char[] alphabet)
183 short[] index = new short[MAX_ASCII + 1];
184 Arrays.fill(index, UNMAPPED);
186 for (char c : alphabet)
188 if (Comparison.isGap(c))
199 * also map lower-case character (unless separately mapped)
201 if (c >= 'A' && c <= 'Z')
203 short lowerCase = (short) (c + ('a' - 'A'));
204 if (index[lowerCase] == UNMAPPED)
206 index[lowerCase] = pos;
215 public String getName()
221 public String getDescription()
227 public boolean isDNA()
233 public boolean isProtein()
239 * Returns a copy of the score matrix as used in getPairwiseScore. If using
240 * this matrix directly, callers <em>must</em> also call
241 * <code>getMatrixIndex</code> in order to get the matrix index for each
242 * character (symbol).
245 * @see #getMatrixIndex(char)
247 public float[][] getMatrix()
249 float[][] v = new float[matrix.length][matrix.length];
250 for (int i = 0; i < matrix.length; i++)
252 v[i] = Arrays.copyOf(matrix[i], matrix[i].length);
258 * Answers the matrix index for a given character, or -1 if unmapped in the
259 * matrix. Use this method only if using <code>getMatrix</code> in order to
260 * compute scores directly (without symbol lookup) for efficiency.
266 public int getMatrixIndex(char c)
268 if (c < symbolIndex.length)
270 return symbolIndex[c];
279 * Answers the matrix index for the gap character, or -1 if unmapped in the
280 * matrix. Use this method only if using <code>getMatrix</code> in order to
281 * compute scores directly (without symbol lookup) for efficiency.
286 public int getGapIndex()
288 return getMatrixIndex(gapCharacter);
292 * Returns the pairwise score for substituting c with d. If either c or d is
293 * an unexpected character, returns 1 for identity (c == d), else the minimum
294 * score value in the matrix.
297 public float getPairwiseScore(char c, char d)
299 if (c >= symbolIndex.length)
301 System.err.println(String.format(BAD_ASCII_ERROR, c));
304 if (d >= symbolIndex.length)
306 System.err.println(String.format(BAD_ASCII_ERROR, d));
310 int cIndex = symbolIndex[c];
311 int dIndex = symbolIndex[d];
312 if (cIndex != UNMAPPED && dIndex != UNMAPPED)
314 return matrix[cIndex][dIndex];
318 * one or both symbols not found in the matrix
320 return c == d ? 1 : getMinimumScore();
324 * pretty print the matrix
327 public String toString()
329 return outputMatrix(false);
333 * Print the score matrix, optionally formatted as html, with the alphabet
334 * symbols as column headings and at the start of each row.
336 * The non-html format should give an output which can be parsed as a score
342 public String outputMatrix(boolean html)
344 StringBuilder sb = new StringBuilder(512);
347 * heading row with alphabet
351 sb.append("<table border=\"1\">");
352 sb.append(html ? "<tr><th></th>" : "");
356 sb.append("ScoreMatrix ").append(getName()).append("\n");
358 for (char sym : symbols)
362 sb.append("<th> ").append(sym).append(" </th>");
366 sb.append("\t").append(sym);
369 sb.append(html ? "</tr>\n" : "\n");
374 for (char c1 : symbols)
378 sb.append("<tr><td>");
380 sb.append(c1).append(html ? "</td>" : "");
381 for (char c2 : symbols)
383 sb.append(html ? "<td>" : "\t")
384 .append(matrix[symbolIndex[c1]][symbolIndex[c2]])
385 .append(html ? "</td>" : "");
387 sb.append(html ? "</tr>\n" : "\n");
391 sb.append("</table>");
393 return sb.toString();
397 * Answers the number of symbols coded for (also equal to the number of rows
398 * and columns of the score matrix)
404 return symbols.length;
408 * Computes an NxN matrix where N is the number of sequences, and entry [i, j]
409 * is sequence[i] pairwise multiplied with sequence[j], as a sum of scores
410 * computed using the current score matrix. For example
412 * <li>Sequences:</li>
417 * <li>Score matrix is BLOSUM62</li>
418 * <li>Gaps treated same as X (unknown)</li>
419 * <li>product [0, 0] = F.F + K.K + L.L = 6 + 5 + 4 = 15</li>
420 * <li>product [1, 1] = R.R + -.- + D.D = 5 + -1 + 6 = 10</li>
421 * <li>product [2, 2] = Q.Q + I.I + A.A = 5 + 4 + 4 = 13</li>
422 * <li>product [3, 3] = G.G + W.W + C.C = 6 + 11 + 9 = 26</li>
423 * <li>product[0, 1] = F.R + K.- + L.D = -3 + -1 + -3 = -8
428 public MatrixI findSimilarities(AlignmentView seqstrings,
429 SimilarityParamsI options)
431 char gapChar = scoreGapAsAny ? (seqstrings.isNa() ? 'N' : 'X')
433 String[] seqs = seqstrings.getSequenceStrings(gapChar);
434 return findSimilarities(seqs, options);
438 * Computes pairwise similarities of a set of sequences using the given
445 protected MatrixI findSimilarities(String[] seqs, SimilarityParamsI params)
447 double[][] values = new double[seqs.length][];
448 for (int row = 0; row < seqs.length; row++)
450 values[row] = new double[seqs.length];
451 for (int col = 0; col < seqs.length; col++)
453 double total = computeSimilarity(seqs[row], seqs[col], params);
454 values[row][col] = total;
457 return new Matrix(values);
461 * Calculates the pairwise similarity of two strings using the given
462 * calculation parameters
469 protected double computeSimilarity(String seq1, String seq2,
470 SimilarityParamsI params)
472 int len1 = seq1.length();
473 int len2 = seq2.length();
476 int width = Math.max(len1, len2);
477 for (int i = 0; i < width; i++)
479 if (i >= len1 || i >= len2)
482 * off the end of one sequence; stop if we are only matching
483 * on the shorter sequence length, else treat as trailing gap
485 if (params.denominateByShortestLength())
491 char c1 = i >= len1 ? gapCharacter : seq1.charAt(i);
492 char c2 = i >= len2 ? gapCharacter : seq2.charAt(i);
493 boolean gap1 = Comparison.isGap(c1);
494 boolean gap2 = Comparison.isGap(c2);
499 * gap-gap: include if options say so, else ignore
501 if (!params.includeGappedColumns())
506 else if (gap1 || gap2)
509 * gap-residue: score if options say so
511 if (!params.includeGaps())
516 float score = getPairwiseScore(c1, c2);
523 * Answers a hashcode computed from the symbol alphabet and the matrix score
527 public int hashCode()
529 int hs = Arrays.hashCode(symbols);
530 for (float[] row : matrix)
532 hs = hs * 31 + Arrays.hashCode(row);
538 * Answers true if the argument is a ScoreMatrix with the same symbol alphabet
539 * and score values, else false
542 public boolean equals(Object obj)
544 if (!(obj instanceof ScoreMatrix))
548 ScoreMatrix sm = (ScoreMatrix) obj;
549 if (Arrays.equals(symbols, sm.symbols)
550 && Arrays.deepEquals(matrix, sm.matrix))
558 * Returns the alphabet the matrix scores for, as a string of characters
564 return new String(symbols);
567 public void setDescription(String desc)
572 public float getMinimumScore()
577 public float getMaximumScore()